There are some questions I have regarding the encoding and decoding of h.264 streams as well as the associated sample projects.
It would appear that the v2.2 SDK contains the only decode example in between the v2.3 and v2.2 SDKs. Was this intentional (i.e. nothing changed for this from v2.2 to v2.3) or was this just a simple mistake?
Regarding the Dx9CUDAGPUDecode project, it would appear the environment variables for this particular project was still pointing to the earlier Cuda 5.0 SDK, and as such anyone downloading the default (i.e. only option on the CUDA members downloads page) 6.0 SDK will have to edit the .vcxproj to adjust for this before it can open properly. Once this is done it compiles and does work indeed. Just curious if this was intentionally left this way or just not on the high totem of things to be updated?
The next question has to do with the ample examples of encoding using various DX versions, but only one decode example using the DX9 library. Are there plans for other examples, or are there no DX call differences between the various versions and as such one only needs to migrate the project to the DX version desired?
The final (more important) question would be: Are there any examples or perhaps pointers on how to build a “.index” file for any .h264 file generated using one of the DX(v#) encode examples? All of the examples appear to generate a .h264 file with no issues at all, however there doesn’t appear to be any that generate the .h264 and .index file (i.e. so one could use the only decode example to verify the encoding worked by playing back the newly generated .h264 file). Any pointers on this would be greatly appreciated.